Els van Meijel

31 Evaluation of the STEPP | Chapter 2 with different positivity thresholds, and thus allows determination of the optimal cut-off point in a sample for a clinically optimal discriminative ability of a test. At the lowest cut-off point, all subjects are classified as test-positive (including the diseased), resulting in 100% sensitivity but 0% specificity. On the other end, at the highest cut- off point, all subjects (including the diseased) are classified as non-diseased, resulting in 0% sensitivity and 100% specificity. The area under the ROC curve (AUC) reflects the overall predictive performance of a test. The maximum value is 1, which means a 100% accurate test, whereas an AUC of 0.50 indicates the test does not perform better than chance. We used the STEPP score as the index test. Diagnosed PTSD and a positive score on the self-report PTSD measures were used as the reference tests. Results of the index test and the reference tests were cross-classified in 2-by-2 tables, and sensitivity, specificity, positive and negative predictive values were calculated. The optimal cut-off score for the STEPP for our sample was based on the decision to maximize both sensitivity and negative predictive values. Intra-rater reliability was tested for the STEPP: the Kappa statistic was used to determine consistency between the first and the second assessment by the same rater. Statistical analyses were performed using SPSS 18 and 19 (IBM Statistical Product and Service Solutions, Chicago, Ill). Results A total of 161 children and 156 parents completed the first assessment within one week of the accident. Demographic, trauma-related and clinical characteristics of this sample are reported in Table 2.1. There were no significant differences between participants and non-participants with regard to age (U = 8170, Z = -.467, p = .64), sex (χ 2 = 1.21, p = .27) or injury severity (U = 5419, Z = -1.367, p = .17). Three months after the accident, 146 children and 139 parents completed the second assessment. Those who dropped out after the first assessment did not differ significantly from those who completed the second assessment in terms of age (U = 908, Z = -.736, p = .46), injury severity (U = 939, Z = -.429, p = .67), sex (χ 2 = .02, p = .88) or STEPP score (t(159) = -1.92, p = .06).